Have you always wondered whether the "free shipping" message or the "Send coupon" message is better, whether to send an email or an SMS? Thanks to the Q Message node, you can perform an A|B test even in automated flows, using different messages and different channels.
Let's start with an example
An eCommerce wants to send a message for the abandoned cart but wants to understand which channel and type of discounting converts more: Newsletter A - with free shipping - or Newsletter B - with a 5% discount, additionally, they would like to figure out whether it is better to send an email or an SMS.
In this case, we can use the new Q Message node.
In the same node, you can configure up to 5 different messages, both in content and channel, so there will be only one exit.
The automated flow self-evaluates and will progressively send the message that performs better.
How it works
At the initial stage, the node divides incoming sessions in a random but evenly distributed manner between the exits (e.g., if the node has four exits, sessions will pass at 25% from each exit).
After a sufficient number of sessions, it begins to evaluate how the automated flow is going, the percentages with which the sessions are distributed between the exits are gradually modified to converge the majority of sessions towards the exit that maximizes the objective.
How to configure it
The Q Message node is designed to be simple and intuitive to configure:
- Number of messages: how many messages do you want to evaluate? You can create up to 5 different messages.
- Recalculate exits: Choose based on what you want to evaluate the automated flow. The objective can be: flow success or Number of conversions.
- By flow success, we mean how many sessions have ended exiting from an exit node with a positive outcome. By number of conversions, we mean the conversions of goals that were made by contacts starting from the communications sent by the automated flow.
- Conversion goal: if you chose to evaluate the flow success based on the number of conversions, you must choose which goal to evaluate.
Returning to our example
If we return to the example described earlier, we set:
-
- Number of messages: 4 corresponding to the two newsletters A and B and the messages A and B.
- Recalculate outputs: we set the number of conversions, since we want to know which of the two communications generates more conversions, hence purchases.
- Conversion goal: we choose the goal representing purchases.
Initially, the system will distribute the four outputs equally, sending each newsletter to 25% of the contacts. Then, after an adequate amount of time to evaluate the results, the system will begin to send the message that generates the highest number of conversions to a higher percentage of contacts.
Over time
The Q Message node will continue to learn and adapt to the performance of the two communications, automatically adjusting the distribution between the outputs of the node.
You will see at any time, inside the node, the current distribution percentages.